Case study: how a global bank built its climate risk framework — from TCFD alignment to regulatory stress testing

Why It Matters

When the European Central Bank published results from its first economy-wide climate stress test in 2022, it found that 60 percent of eurozone banks had no internal climate risk framework at all (ECB, 2022). By 2025, the landscape had shifted dramatically: the Network for Greening the Financial System reported that 46 central banks and supervisors across every major economy had either completed or mandated climate stress tests, and the Bank of England's Climate Biennial Exploratory Scenario revealed that UK banks could face cumulative credit losses of up to £225 billion under a late-action transition pathway (NGFS, 2025). For global banks holding trillions in climate-exposed assets, building a robust climate risk framework is no longer an optional sustainability exercise. It is a regulatory imperative with direct implications for capital adequacy, credit provisioning, and long-term franchise value.

This case study traces how a composite global bank, drawing on publicly disclosed experiences from institutions including HSBC, ING, and BNP Paribas, built its climate risk architecture from initial TCFD alignment through to its first supervisory stress test submission.

Key Concepts

TCFD and ISSB alignment. The Task Force on Climate-related Financial Disclosures established the four-pillar framework of Governance, Strategy, Risk Management, and Metrics and Targets that most banks use as a starting point. Since 2024, the International Sustainability Standards Board's IFRS S2 standard has superseded TCFD as the global baseline for climate disclosure, requiring scenario analysis across multiple time horizons and granular Scope 3 financed-emissions data (ISSB, 2024).

Transition risk vs. physical risk. Transition risk captures the financial impact of policy changes, technology shifts, and market repricing on a bank's loan book and trading positions. Physical risk captures acute events such as floods and chronic shifts such as rising sea levels that impair collateral values or borrower creditworthiness. Most frameworks must model both categories across short (one to three years), medium (three to ten years), and long (ten to thirty years plus) horizons.

Climate scenario analysis. Supervisors require banks to model portfolios against standardized scenarios, typically derived from the NGFS framework. The most common set includes an orderly transition (Net Zero 2050), a disorderly transition (Delayed Transition), and a hot-house world (Current Policies). Each scenario embeds assumptions about carbon prices, GDP trajectories, energy mixes, and physical climate parameters sourced from integrated assessment models.

Financed emissions and portfolio alignment. Under the Partnership for Carbon Accounting Financials (PCAF) methodology, banks quantify the greenhouse gas emissions attributable to their lending, investment, and underwriting activities. The PCAF Global Core Team reported that over 500 financial institutions with combined assets exceeding $90 trillion had adopted PCAF by late 2025, up from 340 at the end of 2023 (PCAF, 2025). Portfolio alignment metrics such as the Implied Temperature Rise score translate these emissions into a forward-looking measure of how closely a bank's portfolio tracks a given warming pathway.

Supervisory stress testing. Regulators including the ECB, the Bank of England, and the Federal Reserve have moved from exploratory exercises to binding supervisory stress tests. In the ECB's 2024 cycle, banks were required to quantify projected credit losses, market value changes, and operational disruptions under each NGFS scenario across a thirty-year horizon (ECB, 2024). Results now feed into the Supervisory Review and Evaluation Process, meaning poor performance can trigger higher capital requirements.

What's Working

Governance integration is maturing. Leading banks have moved climate risk oversight from standalone sustainability committees to board-level risk committees with explicit mandates. HSBC, for example, restructured its Group Risk Committee in 2024 to include a dedicated climate risk sub-committee chaired by a non-executive director with quantitative risk expertise (HSBC, 2024). This integration ensures that climate risk receives the same governance discipline as credit, market, and operational risk.

Centralized climate data platforms are reducing friction. ING developed a centralized Climate Risk Data Hub in 2023 that aggregates counterparty-level emissions data, physical risk scores, and transition pathway indicators into a single platform accessible to front-office relationship managers and risk analysts alike (ING, 2024). By connecting PCAF-compliant emissions calculations with scenario outputs, the hub reduced the time required to complete counterparty-level climate risk assessments from twelve weeks to under three weeks.

Sector-level transition pathway analysis is improving credit decisions. Banks that have built granular sector transition models report better-informed credit origination. BNP Paribas published sector-specific transition pathways covering oil and gas, power generation, automotive, steel, cement, and aviation, each with milestones and performance indicators benchmarked against IEA Net Zero 2050 (BNP Paribas, 2025). Relationship managers use these pathways to evaluate whether a borrower's capital expenditure plan is consistent with a credible decarbonization trajectory.

Regulatory engagement is yielding clearer methodological guidance. The ECB's 2024 Good Practices Report identified eleven methodological areas where bank practices had converged, including the use of NGFS Phase IV scenarios, the treatment of carbon pricing in probability-of-default models, and approaches to physical risk collateral haircuts (ECB, 2024). Banks that participated actively in supervisory dialogue reported fewer surprises during the formal stress test process.

What Isn't Working

Data quality remains the single largest bottleneck. Despite progress in centralized platforms, the majority of counterparty-level emissions data is still estimated rather than reported. The PCAF Global Core Team acknowledged in its 2025 annual report that over 70 percent of Scope 3 financed-emissions calculations rely on sector-average or spend-based proxies rather than primary company data (PCAF, 2025). This imprecision cascades through scenario models and produces wide uncertainty bands in stress test outputs that limit their usefulness for capital allocation decisions.

Physical risk modeling lags transition risk. Most banks have invested heavily in transition risk models, which rely on economic variables and policy assumptions. Physical risk assessment, by contrast, requires geospatial data at the asset level, which is often unavailable for large commercial loan portfolios. The Bank of England's 2025 climate review noted that fewer than 30 percent of UK-regulated banks had achieved property-level geocoding for more than half of their real estate-secured lending (BoE, 2025).

Time horizon mismatches create governance tension. Climate scenarios typically run thirty years or longer, but most bank risk appetite frameworks and regulatory capital calculations operate on a one-to-five-year horizon. Risk committees struggle to translate a projected 2050 credit loss into a 2026 provisioning decision, and boards question the credibility of models extrapolating across three decades of technological and political uncertainty.

Internal incentive structures have not caught up. Front-office incentives remain anchored to short-term revenue and volume targets. Climate risk teams report difficulty influencing credit approval decisions when a deal is profitable under current conditions but exposed under transition scenarios. Only a handful of banks, including Standard Chartered and Barclays, have linked a portion of senior executive compensation to climate-related KPIs (CDP, 2025).

Key Players

Established Leaders

• HSBC — Pioneered board-level climate risk governance restructuring; over $750 billion in sustainable finance facilitated since 2020.
• BNP Paribas — Published sector transition pathways across six high-emitting industries; leading PCAF adopter in Europe.
• ING — Developed the Terra approach to portfolio alignment and built a centralized Climate Risk Data Hub serving 40,000 corporate clients.
• Standard Chartered — Integrated climate risk into performance scorecards and compensation; active in emerging market transition finance.

Emerging Startups

• Cervest — AI-driven physical climate risk analytics platform providing asset-level risk scores to banks and insurers.
• Risilience — Enterprise climate scenario modeling software used by several FTSE 100 financial institutions.
• OS-Climate — Open-source climate data and analytics platform backed by the Linux Foundation to standardize financed-emissions calculations.
• Intensel — Granular physical risk intelligence for real estate and infrastructure portfolios.

Key Investors/Funders

• Glasgow Financial Alliance for Net Zero (GFANZ) — Convenes over 675 financial institutions with $130 trillion in assets committed to net-zero portfolios.
• Bezos Earth Fund — Provided $100 million to support OS-Climate and open-source climate data infrastructure.
• ClimateArc (UK Government) — Public-private climate risk research funding initiative launched in 2025 to accelerate physical risk data availability.

Real-World Examples

HSBC's supervisory stress test journey. HSBC disclosed in its 2024 Annual Report that it had completed the ECB's climate stress test cycle across its eurozone subsidiary, modeling transition and physical risk impacts on a €200 billion loan book. The bank invested over £40 million in climate data infrastructure, hired 85 specialist climate risk analysts, and developed bespoke probability-of-default overlays for carbon-intensive sectors. The exercise revealed that under a delayed transition scenario, expected credit losses in its oil and gas portfolio could increase by 18 percent over a ten-year horizon. HSBC subsequently tightened lending criteria for upstream fossil fuel expansion and introduced a mandatory climate risk assessment for all exposures above £10 million (HSBC, 2024).

ING's Terra portfolio alignment. ING's Terra approach, launched in 2019 and updated annually, measures the alignment of nine carbon-intensive sectors against IEA scenarios. By 2025, the bank reported that its power generation portfolio had moved from a 2.7°C implied temperature pathway to 2.1°C, driven by a deliberate shift in lending toward renewable energy projects and engagement with coal-dependent utilities on transition plans (ING, 2024). The approach demonstrated that portfolio steering could produce measurable alignment shifts within five years, though ING acknowledged that sectors such as shipping and aviation remained stubbornly above 2.5°C.

Bank of England's Climate Biennial Exploratory Scenario. The BoE's CBES, completed in its second round in 2025, stress-tested the UK's seven largest banks and several major insurers against three NGFS scenarios over a thirty-year horizon. Results showed that aggregate banking sector losses ranged from £80 billion under an early orderly transition to £225 billion under a late disorderly scenario. The BoE used findings to issue targeted supervisory guidance, requiring banks with the weakest physical risk modeling capabilities to submit remediation plans within twelve months (BoE, 2025).

BNP Paribas sector pathway integration. BNP Paribas embedded its published sector transition pathways directly into its credit approval workflow in 2025. Relationship managers are now required to assess each borrower against the relevant sector pathway and flag deviations for escalated review. In the first year of implementation, the bank reported that 12 percent of proposed exposures in high-emitting sectors were either restructured with climate covenants or declined, resulting in a 9 percent reduction in financed emissions intensity across its energy and industrial portfolio (BNP Paribas, 2025).

Action Checklist

• Establish board-level climate risk governance. Create a dedicated climate risk sub-committee under the main risk committee with clear escalation authority and reporting lines to the full board.
• Build a centralized climate data infrastructure. Aggregate counterparty emissions, physical risk scores, and scenario outputs into a single platform accessible to risk, credit, and front-office teams.
• Adopt PCAF for financed emissions. Measure and disclose portfolio-level emissions using the PCAF methodology; prioritize primary data collection over sector-average proxies.
• Develop sector-specific transition pathways. Benchmark borrower decarbonization trajectories against credible scenarios (IEA, NGFS) and embed pathway assessments into credit origination and review processes.
• Invest in physical risk geocoding. Achieve property-level geocoding for at least 80 percent of real estate-secured lending and overlay hazard data from providers such as Cervest or Intensel.
• Align incentives with climate risk outcomes. Link a portion of senior executive and relationship manager compensation to measurable climate risk KPIs including financed-emissions intensity reductions.
• Engage proactively with supervisors. Participate in industry working groups, respond to consultation papers, and maintain an open dialogue with prudential regulators on evolving methodological expectations.

FAQ

How long does it take to build a climate risk framework from scratch? Most global banks report a three-to-five-year journey from initial TCFD gap assessment to full integration with supervisory stress testing workflows. The first year typically focuses on governance design and data architecture; years two and three cover scenario model development and pilot testing; and years four and five embed outputs into credit decisions, capital planning, and disclosure. Banks that invested early, such as HSBC and ING, are now in their second or third iteration cycle.

What is the biggest obstacle to accurate climate stress testing? Data quality consistently ranks as the primary challenge. Over 70 percent of financed-emissions calculations rely on estimated rather than reported data (PCAF, 2025), and physical risk assessments require asset-level geocoding that most banks lack for large commercial portfolios. Supervisors have responded by allowing phased data quality improvements, but banks that depend heavily on proxies produce stress test outputs with wide uncertainty ranges that are difficult to translate into concrete capital or provisioning actions.

Do climate stress tests affect capital requirements? In the eurozone, ECB climate stress test results feed into the Supervisory Review and Evaluation Process and can influence Pillar 2 capital guidance, though they are not yet incorporated as a direct Pillar 1 capital charge. The Bank of England has taken a similar approach, using CBES results to set supervisory expectations rather than prescriptive capital add-ons. However, regulators have signaled that binding capital implications could follow as methodologies mature, and the Basel Committee's 2025 consultation on climate-related financial risks proposed principles for integrating climate into Pillar 1 frameworks.

How should banks handle the thirty-year time horizon in climate scenarios? Leading practice involves translating long-horizon scenario outputs into medium-term risk appetite metrics and short-term portfolio steering actions. For example, a projected 2040 credit loss concentration in carbon-intensive sectors can inform a 2026 sector exposure limit or trigger enhanced due diligence requirements. Banks such as BNP Paribas use interim milestones at five-year intervals to bridge the gap between long-run model outputs and near-term credit and capital decisions.

What role do fintech and data providers play? Specialized climate analytics firms have become essential partners. Platforms like Cervest and Intensel provide the granular physical risk data that banks cannot generate internally, while OS-Climate's open-source tools offer standardized emissions calculation engines. Banks increasingly treat these providers as critical infrastructure rather than discretionary vendors, embedding their outputs directly into risk systems and supervisory reporting pipelines.